Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x281595cba90>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x28159d12ba8>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.0.0
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function

    inputs_real=tf.placeholder(tf.float32,[None,image_width,image_height,image_channels],name='inputs_real')
    inputs_z = tf.placeholder(tf.float32,[None,z_dim],name='inputs_z')
    learning_rate=tf.placeholder(tf.float32,name="l_rate")
    return inputs_real,inputs_z,learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param image: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    #input is 28x28x3
    alpha=.2
    
    with tf.variable_scope('discriminator',reuse=reuse):
        
        x1=tf.layers.conv2d(images,64,5,strides=2,padding='same',activation=None)
        x1=tf.maximum(alpha*x1,x1)
        # 14x14x64
        
        x2=tf.layers.conv2d(x1,128,5,strides=2,padding='same',activation=None)
        x2=tf.layers.batch_normalization(x2,training=True)
        x2=tf.maximum(alpha*x2,x2)
        # 7x7x128

        x3=tf.layers.conv2d(x2,256,5,strides=2,padding='same',activation=None)
        x3=tf.layers.batch_normalization(x3,training=True)
        x3=tf.maximum(alpha*x3,x3)
        # 3x3x256

        x4=tf.layers.conv2d(x3,512,5,strides=2,padding='same',activation=None)
        x4=tf.layers.batch_normalization(x4,training=True)
        x4=tf.maximum(alpha*x4,x4)

        x3_flat=tf.reshape(x4,(-1,2*2*512))
        logits=tf.layers.dense(x3_flat,1)
        out=tf.nn.sigmoid(logits)
        
    return out,logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    alpha=.2
    with tf.variable_scope('generator',reuse= not is_train):
        x1=tf.layers.dense(z,2*2*512)
        # 2048
        # print(x1.shape)
    
        x1 = tf.reshape(x1, (-1, 2,2, 512))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        x1 = tf.layers.dropout(x1,0.6)
        # print(x1.shape)
        # 2x2x512 now
        
        x2 = tf.layers.conv2d_transpose(x1, 256, 5, strides=2, padding='valid')
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(alpha * x2, x2)
        # print(x2.shape)
        # 7x7x256 now
        
        x3 = tf.layers.conv2d_transpose(x2, 128, 5, strides=2, padding='same')
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        # 14x14x128
        # print(x3.shape)
        


        # Output layer
        logits = tf.layers.conv2d_transpose(x3, out_channel_dim, 2, strides=2, padding='same')
        # print(logits.shape)
        # 28x28x5 now
        
        out = tf.tanh(logits)
    
    return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    smooth=.1
    
    g_model=generator(input_z,out_channel_dim,is_train=True)
    
    d_model_real,d_logits_real=discriminator(input_real)
    d_model_fake,d_logits_fake=discriminator(g_model,reuse=True)

    d_loss_real=tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real,labels=tf.ones_like(d_logits_real)*(1-smooth)))
    d_loss_fake=tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake,labels=tf.zeros_like(d_logits_real)))
    
    discriminator_loss=d_loss_real+d_loss_fake

    generator_loss=tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake,labels=tf.ones_like(d_logits_fake)))
    
    return discriminator_loss,generator_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    
    trainables=tf.trainable_variables()
    discriminator_vars=[trainable for trainable in trainables if 'discriminator' in trainable.name ]
    generator_vars=[trainable for trainable in trainables if 'generator' in trainable.name ]
    
    update_ops=tf.get_collection(tf.GraphKeys.UPDATE_OPS)
    generator_update_ops=[trainable for trainable in update_ops if 'generator' in trainable.name ]
    discriminator_update_ops=[trainable for trainable in update_ops if 'discriminator' in trainable.name ]
    
    with tf.control_dependencies(discriminator_update_ops):
        d_train_opt=tf.train.AdamOptimizer(learning_rate=learning_rate,beta1=beta1).minimize(d_loss,var_list=discriminator_vars)

    with tf.control_dependencies(generator_update_ops):
        g_train_opt=tf.train.AdamOptimizer(learning_rate=learning_rate,beta1=beta1).minimize(g_loss,var_list=generator_vars)
    
    return d_train_opt, g_train_opt

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    inputs_real,inputs_z,l_rate=model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)
    d_loss,g_loss=model_loss(inputs_real, inputs_z, data_shape[3])
    d_train_opt, g_train_opt=model_opt(d_loss, g_loss, learning_rate, beta1)
    
    step=0
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                step +=1
                # TODO: Train Model
                batch_images = batch_images.reshape(batch_size, data_shape[1], data_shape[2], data_shape[3])
                batch_images = batch_images*2.
                
                # print(z_dim)
                batch_z=np.random.uniform(-1,1,size=(batch_size,z_dim))

                # Run optimizers
                _ = sess.run(d_train_opt, feed_dict={ inputs_real: batch_images,inputs_z: batch_z,l_rate:learning_rate})
                _ = sess.run(g_train_opt, feed_dict={inputs_z: batch_z, l_rate:learning_rate})
                
                
                if step % 10 == 0:
                    train_loss_d = d_loss.eval({inputs_z: batch_z, inputs_real: batch_images})
                    train_loss_g = g_loss.eval({inputs_z: batch_z})

                    print("Epoch {}/{}...".format(epoch_i+1, epoch_count),
                            "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
        

                if step % 100 == 0:
                    _ = show_generator_output(sess, 25, inputs_z, data_shape[3], data_image_mode)
                
            # At the end of each epoch, get the losses and print them out
            train_loss_d = d_loss.eval({inputs_z: batch_z, inputs_real: batch_images})
            train_loss_g = g_loss.eval({inputs_z: batch_z})            
            print("Epoch {}/{}...".format(epoch_i+1, epoch_count),
                      "Discriminator Loss: {:.4f}...".format(train_loss_d),
                      "Generator Loss: {:.4f}".format(train_loss_g))

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 8
z_dim = 100
learning_rate = .001
beta1 = .2

# tf.reset_default_graph()
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 0.9532... Generator Loss: 0.8186
Epoch 1/2... Discriminator Loss: 0.3771... Generator Loss: 3.1303
Epoch 1/2... Discriminator Loss: 1.0840... Generator Loss: 0.6402
Epoch 1/2... Discriminator Loss: 0.9122... Generator Loss: 0.9263
Epoch 1/2... Discriminator Loss: 1.3149... Generator Loss: 0.5091
Epoch 1/2... Discriminator Loss: 1.8948... Generator Loss: 0.8006
Epoch 1/2... Discriminator Loss: 1.4248... Generator Loss: 0.6305
Epoch 1/2... Discriminator Loss: 2.4786... Generator Loss: 4.6117
Epoch 1/2... Discriminator Loss: 1.1471... Generator Loss: 1.3337
Epoch 1/2... Discriminator Loss: 1.6257... Generator Loss: 0.3426

Epoch 1/2... Discriminator Loss: 0.6171... Generator Loss: 1.6628
Epoch 1/2... Discriminator Loss: 0.9937... Generator Loss: 1.0261
Epoch 1/2... Discriminator Loss: 0.7559... Generator Loss: 1.1588
Epoch 1/2... Discriminator Loss: 2.4904... Generator Loss: 0.1349
Epoch 1/2... Discriminator Loss: 1.3570... Generator Loss: 0.6356
Epoch 1/2... Discriminator Loss: 0.9289... Generator Loss: 1.0831
Epoch 1/2... Discriminator Loss: 1.4064... Generator Loss: 0.4867
Epoch 1/2... Discriminator Loss: 1.2534... Generator Loss: 0.7457
Epoch 1/2... Discriminator Loss: 1.0369... Generator Loss: 0.9363
Epoch 1/2... Discriminator Loss: 2.0784... Generator Loss: 0.2085

Epoch 1/2... Discriminator Loss: 2.1960... Generator Loss: 0.2241
Epoch 1/2... Discriminator Loss: 1.4064... Generator Loss: 2.4100
Epoch 1/2... Discriminator Loss: 1.1515... Generator Loss: 0.6064
Epoch 1/2... Discriminator Loss: 1.6492... Generator Loss: 0.7030
Epoch 1/2... Discriminator Loss: 1.0612... Generator Loss: 0.7872
Epoch 1/2... Discriminator Loss: 1.7515... Generator Loss: 0.2926
Epoch 1/2... Discriminator Loss: 1.3639... Generator Loss: 0.5257
Epoch 1/2... Discriminator Loss: 1.2420... Generator Loss: 0.8294
Epoch 1/2... Discriminator Loss: 1.2759... Generator Loss: 0.7582
Epoch 1/2... Discriminator Loss: 1.8210... Generator Loss: 0.2651

Epoch 1/2... Discriminator Loss: 1.5268... Generator Loss: 0.3918
Epoch 1/2... Discriminator Loss: 1.0015... Generator Loss: 1.0098
Epoch 1/2... Discriminator Loss: 1.3500... Generator Loss: 0.8384
Epoch 1/2... Discriminator Loss: 1.5023... Generator Loss: 0.4136
Epoch 1/2... Discriminator Loss: 2.1242... Generator Loss: 0.3673
Epoch 1/2... Discriminator Loss: 2.7188... Generator Loss: 0.1236
Epoch 1/2... Discriminator Loss: 1.2037... Generator Loss: 0.5959
Epoch 1/2... Discriminator Loss: 2.9746... Generator Loss: 0.0811
Epoch 1/2... Discriminator Loss: 1.3704... Generator Loss: 0.4831
Epoch 1/2... Discriminator Loss: 1.6895... Generator Loss: 0.3578

Epoch 1/2... Discriminator Loss: 1.1356... Generator Loss: 1.3328
Epoch 1/2... Discriminator Loss: 1.2632... Generator Loss: 0.7349
Epoch 1/2... Discriminator Loss: 1.7660... Generator Loss: 0.3272
Epoch 1/2... Discriminator Loss: 0.7699... Generator Loss: 1.1401
Epoch 1/2... Discriminator Loss: 1.1086... Generator Loss: 1.1827
Epoch 1/2... Discriminator Loss: 1.3916... Generator Loss: 0.4737
Epoch 1/2... Discriminator Loss: 2.8394... Generator Loss: 0.0983
Epoch 1/2... Discriminator Loss: 1.2528... Generator Loss: 0.8383
Epoch 1/2... Discriminator Loss: 1.4745... Generator Loss: 0.4357
Epoch 1/2... Discriminator Loss: 1.7289... Generator Loss: 0.3186

Epoch 1/2... Discriminator Loss: 0.9334... Generator Loss: 0.9927
Epoch 1/2... Discriminator Loss: 1.3780... Generator Loss: 0.5543
Epoch 1/2... Discriminator Loss: 1.5143... Generator Loss: 0.5592
Epoch 1/2... Discriminator Loss: 0.7741... Generator Loss: 1.4194
Epoch 1/2... Discriminator Loss: 1.4345... Generator Loss: 0.4349
Epoch 1/2... Discriminator Loss: 2.6776... Generator Loss: 0.1034
Epoch 1/2... Discriminator Loss: 1.9706... Generator Loss: 0.2796
Epoch 1/2... Discriminator Loss: 2.5512... Generator Loss: 0.1655
Epoch 1/2... Discriminator Loss: 2.0882... Generator Loss: 0.2641
Epoch 1/2... Discriminator Loss: 0.8488... Generator Loss: 0.9604

Epoch 1/2... Discriminator Loss: 2.0091... Generator Loss: 0.2212
Epoch 1/2... Discriminator Loss: 1.4378... Generator Loss: 0.4458
Epoch 1/2... Discriminator Loss: 1.2266... Generator Loss: 0.6058
Epoch 1/2... Discriminator Loss: 1.4808... Generator Loss: 0.3958
Epoch 1/2... Discriminator Loss: 1.4061... Generator Loss: 0.5718
Epoch 1/2... Discriminator Loss: 3.4621... Generator Loss: 0.0656
Epoch 1/2... Discriminator Loss: 2.5399... Generator Loss: 0.1247
Epoch 1/2... Discriminator Loss: 1.4185... Generator Loss: 0.4388
Epoch 1/2... Discriminator Loss: 1.6476... Generator Loss: 0.3647
Epoch 1/2... Discriminator Loss: 1.8179... Generator Loss: 0.3123

Epoch 1/2... Discriminator Loss: 0.7808... Generator Loss: 1.6336
Epoch 1/2... Discriminator Loss: 0.8015... Generator Loss: 2.6226
Epoch 1/2... Discriminator Loss: 1.5712... Generator Loss: 0.4036
Epoch 1/2... Discriminator Loss: 2.2116... Generator Loss: 0.2176
Epoch 1/2... Discriminator Loss: 1.2390... Generator Loss: 0.5901
Epoch 1/2... Discriminator Loss: 0.9797... Generator Loss: 0.9237
Epoch 1/2... Discriminator Loss: 2.2533... Generator Loss: 0.1756
Epoch 1/2... Discriminator Loss: 2.8217... Generator Loss: 0.1002
Epoch 1/2... Discriminator Loss: 1.8733... Generator Loss: 0.2543
Epoch 1/2... Discriminator Loss: 1.2530... Generator Loss: 0.5587

Epoch 1/2... Discriminator Loss: 1.2272... Generator Loss: 0.5439
Epoch 1/2... Discriminator Loss: 0.6784... Generator Loss: 1.6944
Epoch 1/2... Discriminator Loss: 1.8460... Generator Loss: 0.2926
Epoch 1/2... Discriminator Loss: 1.0511... Generator Loss: 1.0279
Epoch 1/2... Discriminator Loss: 1.3711... Generator Loss: 0.4564
Epoch 1/2... Discriminator Loss: 1.0112... Generator Loss: 1.6721
Epoch 1/2... Discriminator Loss: 1.0930... Generator Loss: 0.6951
Epoch 1/2... Discriminator Loss: 0.8583... Generator Loss: 1.0259
Epoch 1/2... Discriminator Loss: 1.3529... Generator Loss: 0.5940
Epoch 1/2... Discriminator Loss: 1.3536... Generator Loss: 0.6414

Epoch 1/2... Discriminator Loss: 1.0124... Generator Loss: 1.0575
Epoch 1/2... Discriminator Loss: 0.7278... Generator Loss: 1.2569
Epoch 1/2... Discriminator Loss: 1.6574... Generator Loss: 0.3155
Epoch 1/2... Discriminator Loss: 1.0287... Generator Loss: 1.6419
Epoch 1/2... Discriminator Loss: 0.8898... Generator Loss: 1.2581
Epoch 1/2... Discriminator Loss: 2.3787... Generator Loss: 0.1848
Epoch 1/2... Discriminator Loss: 1.3397... Generator Loss: 1.8428
Epoch 1/2... Discriminator Loss: 1.1325... Generator Loss: 0.6893
Epoch 1/2... Discriminator Loss: 2.0296... Generator Loss: 3.8308
Epoch 1/2... Discriminator Loss: 1.8781... Generator Loss: 0.2580

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 8
z_dim = 100
learning_rate = .0001
beta1 = .1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 1

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

100
(?, 28, 28, 3)
28
(?, 100)
inputz1 100
out 3
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)
Epoch 1/1... Discriminator Loss: 1.2189... Generator Loss: 3.6081
Epoch 1/1... Discriminator Loss: 2.5897... Generator Loss: 0.1565
Epoch 1/1... Discriminator Loss: 3.7127... Generator Loss: 0.0602
Epoch 1/1... Discriminator Loss: 2.3352... Generator Loss: 0.2347
Epoch 1/1... Discriminator Loss: 0.8812... Generator Loss: 7.4372
Epoch 1/1... Discriminator Loss: 1.1257... Generator Loss: 7.3123
Epoch 1/1... Discriminator Loss: 0.5116... Generator Loss: 4.1007
Epoch 1/1... Discriminator Loss: 0.4336... Generator Loss: 3.5787
Epoch 1/1... Discriminator Loss: 0.5334... Generator Loss: 5.0651
Epoch 1/1... Discriminator Loss: 0.4439... Generator Loss: 2.6081
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 1.9706... Generator Loss: 0.3274
Epoch 1/1... Discriminator Loss: 0.5528... Generator Loss: 2.4443
Epoch 1/1... Discriminator Loss: 1.1043... Generator Loss: 0.6739
Epoch 1/1... Discriminator Loss: 1.1114... Generator Loss: 0.6460
Epoch 1/1... Discriminator Loss: 0.8193... Generator Loss: 1.2051
Epoch 1/1... Discriminator Loss: 0.9229... Generator Loss: 3.4124
Epoch 1/1... Discriminator Loss: 1.1062... Generator Loss: 4.8738
Epoch 1/1... Discriminator Loss: 0.8592... Generator Loss: 1.4165
Epoch 1/1... Discriminator Loss: 1.1214... Generator Loss: 0.6314
Epoch 1/1... Discriminator Loss: 0.7139... Generator Loss: 2.9763
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 0.9729... Generator Loss: 5.2825
Epoch 1/1... Discriminator Loss: 1.4439... Generator Loss: 0.4198
Epoch 1/1... Discriminator Loss: 1.5286... Generator Loss: 2.2928
Epoch 1/1... Discriminator Loss: 0.8420... Generator Loss: 0.9792
Epoch 1/1... Discriminator Loss: 1.2696... Generator Loss: 0.5376
Epoch 1/1... Discriminator Loss: 1.1872... Generator Loss: 0.8498
Epoch 1/1... Discriminator Loss: 2.4993... Generator Loss: 0.1386
Epoch 1/1... Discriminator Loss: 0.7746... Generator Loss: 1.6120
Epoch 1/1... Discriminator Loss: 1.0944... Generator Loss: 0.7470
Epoch 1/1... Discriminator Loss: 1.0738... Generator Loss: 3.4221
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 1.6312... Generator Loss: 0.4289
Epoch 1/1... Discriminator Loss: 2.4038... Generator Loss: 0.1584
Epoch 1/1... Discriminator Loss: 1.6564... Generator Loss: 0.3491
Epoch 1/1... Discriminator Loss: 1.3014... Generator Loss: 0.5878
Epoch 1/1... Discriminator Loss: 1.2544... Generator Loss: 0.8239
Epoch 1/1... Discriminator Loss: 1.9449... Generator Loss: 0.3678
Epoch 1/1... Discriminator Loss: 1.2784... Generator Loss: 0.9941
Epoch 1/1... Discriminator Loss: 0.9358... Generator Loss: 1.0406
Epoch 1/1... Discriminator Loss: 1.8798... Generator Loss: 0.2877
Epoch 1/1... Discriminator Loss: 1.0914... Generator Loss: 0.7425
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 0.7121... Generator Loss: 2.0084
Epoch 1/1... Discriminator Loss: 1.6317... Generator Loss: 0.3298
Epoch 1/1... Discriminator Loss: 1.7885... Generator Loss: 0.3045
Epoch 1/1... Discriminator Loss: 0.8316... Generator Loss: 0.9907
Epoch 1/1... Discriminator Loss: 0.8674... Generator Loss: 1.0005
Epoch 1/1... Discriminator Loss: 1.0245... Generator Loss: 0.9409
Epoch 1/1... Discriminator Loss: 1.3272... Generator Loss: 0.4950
Epoch 1/1... Discriminator Loss: 0.6441... Generator Loss: 1.9854
Epoch 1/1... Discriminator Loss: 0.9181... Generator Loss: 0.8525
Epoch 1/1... Discriminator Loss: 1.5676... Generator Loss: 0.4539
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 0.9416... Generator Loss: 0.9062
Epoch 1/1... Discriminator Loss: 1.6777... Generator Loss: 0.3644
Epoch 1/1... Discriminator Loss: 1.1056... Generator Loss: 0.9577
Epoch 1/1... Discriminator Loss: 1.3791... Generator Loss: 0.4613
Epoch 1/1... Discriminator Loss: 2.0550... Generator Loss: 0.2418
Epoch 1/1... Discriminator Loss: 1.8310... Generator Loss: 0.2658
Epoch 1/1... Discriminator Loss: 2.1467... Generator Loss: 0.1985
Epoch 1/1... Discriminator Loss: 1.2112... Generator Loss: 0.8619
Epoch 1/1... Discriminator Loss: 1.8066... Generator Loss: 0.3148
Epoch 1/1... Discriminator Loss: 1.3035... Generator Loss: 0.5250
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 2.0354... Generator Loss: 0.2361
Epoch 1/1... Discriminator Loss: 1.6954... Generator Loss: 0.3822
Epoch 1/1... Discriminator Loss: 1.1639... Generator Loss: 0.6030
Epoch 1/1... Discriminator Loss: 1.3417... Generator Loss: 0.5060
Epoch 1/1... Discriminator Loss: 1.5078... Generator Loss: 0.4284
Epoch 1/1... Discriminator Loss: 1.6527... Generator Loss: 0.5181
Epoch 1/1... Discriminator Loss: 1.9963... Generator Loss: 0.2419
Epoch 1/1... Discriminator Loss: 2.1350... Generator Loss: 0.1927
Epoch 1/1... Discriminator Loss: 1.0136... Generator Loss: 0.9971
Epoch 1/1... Discriminator Loss: 2.1225... Generator Loss: 0.2148
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 0.9848... Generator Loss: 0.8615
Epoch 1/1... Discriminator Loss: 1.2004... Generator Loss: 1.0532
Epoch 1/1... Discriminator Loss: 1.3404... Generator Loss: 0.4931
Epoch 1/1... Discriminator Loss: 1.1891... Generator Loss: 0.6602
Epoch 1/1... Discriminator Loss: 2.2407... Generator Loss: 0.1948
Epoch 1/1... Discriminator Loss: 0.7061... Generator Loss: 1.1993
Epoch 1/1... Discriminator Loss: 1.1426... Generator Loss: 0.7165
Epoch 1/1... Discriminator Loss: 1.9622... Generator Loss: 0.2328
Epoch 1/1... Discriminator Loss: 2.0384... Generator Loss: 0.2103
Epoch 1/1... Discriminator Loss: 2.2111... Generator Loss: 0.1816
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 1.0317... Generator Loss: 1.7908
Epoch 1/1... Discriminator Loss: 1.5972... Generator Loss: 0.3598
Epoch 1/1... Discriminator Loss: 1.6584... Generator Loss: 2.9435
Epoch 1/1... Discriminator Loss: 1.1272... Generator Loss: 0.6651
Epoch 1/1... Discriminator Loss: 0.9656... Generator Loss: 1.9873
Epoch 1/1... Discriminator Loss: 0.9495... Generator Loss: 2.1424
Epoch 1/1... Discriminator Loss: 1.5266... Generator Loss: 0.5114
Epoch 1/1... Discriminator Loss: 1.4524... Generator Loss: 0.4276
Epoch 1/1... Discriminator Loss: 1.3731... Generator Loss: 0.4983
Epoch 1/1... Discriminator Loss: 1.2156... Generator Loss: 0.6679
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Epoch 1/1... Discriminator Loss: 1.1343... Generator Loss: 2.0854
Epoch 1/1... Discriminator Loss: 1.5163... Generator Loss: 0.4440
Epoch 1/1... Discriminator Loss: 1.1865... Generator Loss: 0.6218
Epoch 1/1... Discriminator Loss: 1.2588... Generator Loss: 0.5289
Epoch 1/1... Discriminator Loss: 1.4609... Generator Loss: 0.4545
Epoch 1/1... Discriminator Loss: 0.9047... Generator Loss: 1.3969
Epoch 1/1... Discriminator Loss: 1.2033... Generator Loss: 0.6213
Epoch 1/1... Discriminator Loss: 1.2275... Generator Loss: 0.5987
Epoch 1/1... Discriminator Loss: 1.0131... Generator Loss: 0.8830
Epoch 1/1... Discriminator Loss: 0.7257... Generator Loss: 1.1963
(?, 100)
(?, 2048)
(?, 2, 2, 512)
(?, 7, 7, 256)
(?, 14, 14, 128)
(?, 28, 28, 3)

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.